Understanding the mechanical properties of novel UHTCMCs through random forest and regression tree analysis

Publication date: 5 May 2018
Source:Materials & Design, Volume 145
Author(s): Antonio Vinci, Luca Zoli, Diletta Sciti, Cesare Melandri, Stefano Guicciardi
The microstructure and the mechanical properties of ZrB2–SiC–Cf ceramic composites were investigated. The SiC phase, whose amount was varied from 5 to 20vol%, was introduced in order to improve the densification, oxidation resistance and mechanical properties of the composite. The microstructure was analysed by SEM-EDS and image analysis. Increasing the amount of SiC from 5 to 20vol% resulted in an improvement of the materials density, from 90% to 94%. The non-brittle 4-pt flexural strength ranged from 164 to 247MPa, with no clear dependence on the amount of SiC added. The same holds true for the fracture toughness, ranging from 4.8 to 8.4MPa·m^0.5. In order to track the most important microstructural parameters affecting the properties, experimental data were analysed with the Random Forest and Regression Tree statistical models. The statistical analysis demonstrated that among the possible explanatory variables such as porosity, amount of SiC, fibre content, matrix content, SiC/fibre amount ratio, the one having a major influence on both the flexural strength and fracture toughness is the ratio between SiC and carbon fibre content.

Graphical abstract

http://ift.tt/2FCk7i1